To demonstrate the utility of MTAG empirically, we analyze three traits: depressive symptoms (DEP, Neff = 354,862), neuroticism (NEUR, N = 168,105), and subjective well-being (SWB, N = 388,538). Prior GWASs of each of these traits have identified only a handful of loci.7–11 Because of the high genetic correlations between the three traits—in our data, roughly 0.7 in absolute value between each pair—some papers have conducted cross-trait analyses to replicate findings for one of the traits11 or joint meta-analysis to identify new loci.5 We apply MTAG to these traits because we expected the gains in power would be substantial, violations of MTAG’s assumptions would be limited, and the substantive results would be of interest.
